Teams battles bots, Bioshocking AI browser guardrails, Fortibleed fuels ransomware

Teams cracks down on meeting bots, AI guardrails get bypassed, FortiBleed fuels ransomware, and Nissan confirms PeopleSoft breach

 

Microsoft rolls out a new Teams admin policy, “Manage External Bots and Their Access to Meetings,” to detect third‑party bots, hold them in the lobby with labels, and require organizer approval, with future allow lists, full blocks, reports, and audit logs planned.

 

Anthropic’s Fable 5 returns globally after U.S. export controls are lifted, though higher‑risk requests may be routed to weaker models and Mythos restrictions remain, with Commerce reserving the right to reimpose controls.

 

Researchers describe “Bioshocking,” tricking AI browsers into abandoning guardrails via delusional puzzle prompts, while Adversa AI’s “Guardfall” shows how Bash text rewriting can bypass command filters in many coding agents.

 

SOC Radar links FortiBleed credential theft to InkRansom and Lynx ransomware activity across hundreds of FortiGate portals. Nissan confirms employee data theft tied to a PeopleSoft zero‑day campaign linked to ShinyHunters.

 

Read the original article:

Teams battles bots, Bioshocking AI browser guardrails, Fortibleed fuels ransomware